Abstract
Background
The risk of cardiovascular events (CVEs) in peritoneal dialysis (PD) patients is high, but varies widely among individuals. Metabolic acidosis is prevalent in PD patients and may be involved in the development of CVEs. The aim of the study was to evaluate serum bicarbonate as a risk factor and derive a model of new CVE.
Methods
A predictive model was established by performing an observational study in 187 PD patients obtained from the First Affiliated Hospital of Anhui Medical University. The variables were extracted using least absolute shrinkage and selection operator (LASSO) regression, and the modeling was developed using multivariable Cox regression.
Results
Left ventricular hypertrophy (HR = 1.965, 95%CI 1.086–3.557) and history of CVEs (HR = 2.435, 95%CI 1.342–4.49) were risk parameters for a new CVE. Serum albumin (HR = 0.924, 95%CI 0.864–0.989) and bicarbonate levels (HR = 0.817, 95%CI 0.689–0.969) were protective parameters, in which the risk of CVEs was reduced by 7.6% and 18.3% for each 1-unit increase in serum albumin (g/L) and bicarbonate (mmol/L) levels, respectively. A nomogram based on the above predictive indicators was proposed with a C-statistic of 0.806, indicating good discrimination. Moreover, it successfully stratified patients into low-, intermediate-, and high-risk groups.
Conclusions:
We performed a risk prediction model for the development of CVEs in patients with PD, which may help physicians to evaluate the risk of new CVEs and provide a scientific basis for further interventions. Further studies are needed to externally validate current risk models before clinical application.
Keywords: Peritoneal dialysis, serum bicarbonate, cardiovascular events, nomogram, prognosis
1. Introduction
Chronic kidney disease (CKD) has increasingly received attention as a global public health problem, with a population prevalence rate of up to 14.3%, of which peritoneal dialysis (PD) is an important modality of replacement therapy for end-stage kidney disease (ESKD) [1]. Although PD has little impact on hemodynamics, mortality risk in PD population still remains about 6.1–16 times higher than that of the general population, with cardiovascular events (CVEs) being the leading cause of death [2]. Metabolic acidosis (MA), mainly characterized by low serum bicarbonate concentration, is often encountered in patients with ESKD, and is considered to be an important cause of many deleterious clinical effects, including protein-energy wasting, inflammation, endocrine dysfunction, and decreased nutritional status, etc. [3]. Moreover, further study has confirmed that a low time-averaged serum bicarbonate (TA-Bic) level is an independent risk factor for cardiovascular mortality in PD patients [4]. However, the relationship with CVEs has yet not been established so far.
Several risk factors were reported to be associated with CVEs. The Framingham risk score is one of the most widely used tools to assess the individual risk of CVEs in asymptomatic adults, which has been validated in racially diverse general populations [5]. Nevertheless, this tool did not appear to be applicable to patients with CKD, especially those on dialysis. Huang et al. observed 201 hemodialysis patients and evaluated the performance, indicating that the high-risk (>20% 10-year risk) classified by Framingham risk score cannot predict cardiovascular mortality [6].
Only limited data on cardiovascular risk predictive instrument in dialysis patients are available. An observational study published recently by our research group showed that the elderly, history of CVEs, alkaline phosphatase, serum albumin, and culture-positive could predict CVEs risk in patients with PD associated peritonitis [7]. However, unlike general PD patients, patients with PD associated peritonitis had faster loss of residual kidney function, volume imbalance, and greater loss of albumin. In the current study, the objective is to clarify the association between TA-Bic levels and CVEs risk and to develop a simple and effective predictive tool to predict the risk of CVEs in patients starting PD therapy.
2. Methods
2.1. Study population
We consecutively filtered out 220 patients catheterized at our PD center and treated with continuous ambulatory peritoneal dialysis (CAPD) during the period between 1 June 2018 and 1 June 2022. All patients were treated with standard lactate peritoneal dialysate, containing 15 or 25 g of glucose, 5.38 g of sodium chloride, 0.26 g of calcium chloride, 0.051 g of magnesium chloride, and 4.48 g of sodium lactate per 1000 mL (sodium ion: 132 mmol/L, chloride ion: 96 mmol/L, calcium ion: 1.77 mmol/L, magnesium ion: 0.25 mmol/L, and lactate: 40 mmol/L). The patients with age <18 or >80 years, PD for <3 months, history of hemodialysis or kidney transplant before the initiation of PD, recovered renal function, or lost follow-up were excluded. Furthermore, patients refused to give written consents were excluded. This study was performed in accordance with the Declaration of Helsinki and was approved by the Research Ethics Committee of the First Affiliated Hospital, Anhui Medical University (Ethics approval number: PJ2023-11-52). All participants involved in this study provided written informed consent.
2.2. Data collection
Baseline clinical and laboratory data were obtained using the standardized forms. Following indicators were extracted: the demographic variables, including age, gender, and smoking status; physical examination variables consisted of systolic blood pressure (SBP) and diastolic blood pressure (DBP), which were measured twice with a mercury sphygmomanometer after half an hour of rest on admission, and the average was taken, height, and body weight with dry abdomen; health history, including diabetes, history of CVEs, left ventricular hypertrophy (LVH), and peritonitis, which were collected from electronic medical records in our hospital; medication information, including angiotensin-converting enzyme inhibitors or angiotensin-receptor blockers (ACEI or ARBs), alpha blockers, beta blockers, and calcium channel blockers; and laboratory variables from serum, urine, and PD fluids, including serum bicarbonate concentrations, white blood cell count, residual glomerular filtration rate (rGFR), hemoglobin, serum albumin, creatinine, uric acid, C-reactive protein, ferritin, alkaline phosphatase, total cholesterol, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglycerides, calcium, phosphorus, intact parathyroid hormone (iPTH), dialysate/plasma creatinine ratio (D/P Cr) at 4 h, 24-h urine volume, 24-h ultrafiltration volume, total creatinine clearance rate (CCr), and total urea clearance index (Kt/V), were measured on automated chemical analyzers using standard laboratory methods at 3 months of initial PD. On the basis of weight and height, body mass index was calculated. A history of coronary heart disease, heart failure, stroke, and peripheral arterial disease were considered as a history of CVEs [8,9]. In China, we often depend on voltage criteria for interpreting the electrocardiographic-LVH (ECG-LVH) in the clinical practice. The Sokolow-Lyon voltage (SV1 + RV5/6) > 35 mm and the Cornell voltage criterion-based LVH are two such criteria commonly used. The latter is defined as R in aVL + SV3 ≥ 28mm for men and S in V3 + R in aVL > 20 mm for women [10,11]. Serum bicarbonate level measured at every 3 months during the follow-up was calculated as TA-Bic (time averaged value = (N1 + N2 + N3 + … + Nn)/n) [12]. Supplementary Figure 1 outlines the serial measurement of serum bicarbonate during follow-up for one of the patients in this study. Residual kidney function, expressed as rGFR, was calculated by the following formula: RKF = 1/2[UrineCr(μmol/L)/SerumCr(μmol/L) + UrineUrea(mmol/L)/SerumUrea(mmol/L)] × UrineVolume(mL)/1440 [13], where these values were obtained by peritoneal equilibration test (PET). The patients should be treated with standard CAPD before PET, and peritoneal dialysate should be left in the abdomen at night for 8–12 h.
2.3. Procedures and assessment methods
On 1 October 2023, we retrospectively collected baseline and follow-up data of the selected PD patients from electronic medical records in our hospital, and then assessed their outcome events.
2.4. Outcome
The primary outcome of this observational study was a CVE, which was defined as a combination of hospitalizations due to coronary heart disease, congestive heart failure, cerebrovascular accident (including ischemic stroke and cerebral hemorrhage), peripheral arterial disease, and death associated with aneurysm dissection or rupture, cardiovascular procedure, or fatal pulmonary embolism. Heart failure means patients with a reduced ejection fraction (LVEF <40%), diagnosed by echocardiography. Information on CVEs was extracted from electronic medical records in our hospital. All patients were followed up until CVE, transfer to hemodialysis or kidney transplant, or end of the study period (1 June 2023).
2.5. Statistical analysis
Data were expressed as the means ± standard deviations (SDs), median (interquartile ranges), and frequencies (percentages). All continuous variables were tested for normality using skewness and kurtosis tests. Continuous data with a normal distribution were compared by t-test or analysis of variance; skewed continuous data were compared by the Mann–Whitney U or Kruskal–Wallis test. Additionally, count data were analyzed with the Chi-square or Fisher’s exact probability method test.
Subsequently, patients were classified according to TA-Bic (mmol/L) tertiles: <23.5, lower tertile; 23.5–25.1, middle tertile; and >25.1, upper tertile. To identify relevant stress predictors, we used an innovative analytical approach, the least absolute shrinkage and selection operator (LASSO) regression method, which avoids the overfitting of standard regression models by performing conditioning and selection. Importantly, this process allows us to identify key risk and protective factors, while targeting variables that are critical to CVEs and indicating where prevention and intervention should start. Cox regression was then performed to build the final model. The variance inflation factor (vif) was calculated for each variable to assess the presence of covariance in the final model. We performed concordance index (C-index) and time-dependent receiver operating characteristic (ROC) analysis at different cut off times to measure the discrimination and accuracy of the predictive model. Segmentation was subsequently conducted using the tertiles determined by the total score of the nomogram and patients were categorized into low-, intermediate-, and high-risk groups. The cumulative survival rates of the three groups were compared by the Kaplan–Meier method. Additionally, a bootstrap with 1000 resamples was used to plot calibration curve. Finally, we further performed the sensitivity analysis and internal validation to verify the robustness of the prediction model. All data analyses were conducted using SPSS version 25.0 (SPSS, Inc., Chicago, IL), and R software version 4.1.3 (R Foundation for Statistical Computing, Vienna, Austria). A value of p < .05 was considered to be statistically significant.
3. Results
3.1. Baseline characteristics of study participants
Overall 187 eligible PD patients were enrolled in this study (Figure 1). The mean age of 187 patients was 49 years with a median follow-up of 28 months, of whom 55% were female, 26.7% had a history of CVEs. During the follow-up time, a total of 59 patients had CVEs. Baseline serum bicarbonate was <22 mmol/L in 40% of PD patients. Although we found that oral sodium bicarbonate was given to improve acidosis in nearly one-third of PD patients at follow-up, the time-averaged bicarbonate level in our study was slightly lower than those reported for patients in Europe and the US, which may in part stem from dietary differences. Table 1 shows the demographic characteristics, comorbidities, medications, laboratory data, dialysis characteristics, and outcomes of the patients. Supplementary Figure 2 depicts the distribution of TA-Bic in our cohort of PD patients.
Figure 1.
Enrollment and outcomes of the cohort. PD: peritoneal dialysis.
Table 1.
Baseline characteristics of the study populations.
| Variables | Total (n = 187) | A (n = 128) | B (n = 59) | p Value |
|---|---|---|---|---|
| Age (years) | 49 (38, 57) | 48 (36, 56) | 52 (42, 59) | .031 |
| Female (n (%)) | 103 (55.1) | 71 (55.5) | 32 (54.2) | .875 |
| BMI (kg/m2) | 21.48 (19.72, 23.81) | 21.48 (19.75, 23.50) | 22.23 (19.66, 24.38) | .527 |
| Smoking (n (%)) | 56 (29.9) | 32 (25.0) | 24 (40.7) | .030 |
| Diabetes (n (%)) | 30 (16.0) | 16 (12.5) | 14 (23.7) | .052 |
| Left ventricular hypertrophy (n (%)) | 78 (41.7) | 38 (29.7) | 40(67.8) | <.001 |
| History of CVEs (n (%)) | 50 (26.7) | 16 (12.5) | 34 (57.6) | <.001 |
| Incidence of peritonitis (n (%)) | 45 (24.1) | 28 (21.9) | 17 (28.8) | .302 |
| Systolic blood pressure (mmHg) | 139.09 ± 16.16 | 137.34 ± 16.40 | 142.86 ± 15.08 | .030 |
| Diastolic blood pressure (mmHg) | 89.76 ± 11.49 | 89.39 ± 11.79 | 90.56 ± 10.85 | .519 |
| ACEI/ARBs (n (%)) | 37 (19.8) | 23 (18.0) | 14 (23.7) | .358 |
| Alpha blockers (n (%)) | 21 (11.2) | 13 (10.2) | 8 (13.6) | .493 |
| Beta blockers (n (%)) | 78 (41.7) | 58 (45.3) | 20 (33.9) | .141 |
| Calcium channel blockers (n (%)) | 155 (82.9) | 101 (78.9) | 54 (91.5) | .033 |
| TA-Bic (mmol/L) | 24.40 (23.19, 25.55) | 25.00 (23.55, 26.35) | 23.40 (22.75, 24.05) | <.001 |
| Baseline serum bicarbonate (mmol/L) | 22.50 (21.40, 24.10) | 23.53 (21.80, 25.15) | 21.70 (20.55, 22.20) | <.001 |
| Creatinine (μmol/L) | 706.00 (601.20, 855.80) | 701.45 (596.60, 836.05) | 708.00 (616.50, 872.50) | .271 |
| Uric acid (μmol/L) | 438 (382, 486) | 441 (382, 483) | 420 (385, 488) | .599 |
| rGFR (mL/min/1.73 m2) | 3.65 (2.47, 5.15) | 3.77 (2.68, 5.30) | 3.23 (1.90, 4.42) | .089 |
| White blood cell count (×109/L) | 5.92 (4.85, 7.18) | 5.95 (4.84, 7.14) | 5.77 (4.97, 7.24) | .721 |
| CRP (mg/L) | 1.35 (0.70, 3.74) | 1.28 (0.70, 2.60) | 1.40 (0.81, 5.28) | .161 |
| Serum ferritin (μg/L) | 153 (80, 296) | 141 (73, 276) | 189 (98, 318) | .346 |
| Hemoglobin (g/L) | 106.71 ± 18.84 | 108.84 ± 19.18 | 102.08 ± 17.34 | .022 |
| Albumin (g/L) | 36.27 ± 4.62 | 38.01 ± 4.06 | 32.48 ± 3.30 | <.001 |
| Alkaline phosphatase (U/L) | 78.0 (62.0, 102.0) | 72.5 (59.5, 89.5) | 97.0 (75.0, 125.5) | <.001 |
| Total cholesterol (mmol/L) | 4.42 (3.89, 5.13) | 4.42 (3.87, 5.15) | 4.42 (3.97, 5.01) | .729 |
| Triglyceride (mmol/L) | 1.32 (0.94, 1.94) | 1.30 (0.91, 1.98) | 1.33 (0.98, 1.89) | .980 |
| LDL-C (mmol/L) | 2.73 (2.18, 3.33) | 2.73 (2.09, 3.33) | 2.73 (2.31, 3.34) | .635 |
| HDL-C (mmol/L) | 1.15 (0.98, 1.45) | 1.17 (0.98, 1.45) | 1.14 (0.99, 1.45) | .855 |
| Calcium (mmol/L) | 2.22 ± 0.17 | 2.22 ± 0.16 | 2.24 ± 0.19 | .463 |
| Phosphorus (mmol/L) | 1.62 (1.33, 1.92) | 1.66 (1.34, 1.90) | 1.59 (1.33, 2.05) | .949 |
| iPTH (pg/mL) | 235 (100, 347) | 241.0 (128.5, 358.5) | 166.0 (68.7, 334.5) | .078 |
| D/P Cr at 4 h | 0.61 ± 0.13 | 0.60 ± 0.13 | 0.64 ± 0.14 | .108 |
| 24-h residual urine volume (L) | 0.80 (0.50, 1.10) | 0.80 (0.50, 1.10) | 0.80 (0.56, 1.10) | .962 |
| 24-h ultrafiltration volume (L) | 0.46 (0.09, 0.70) | 0.49 (0.11, 0.70) | 0.41 (0.00, 0.69) | .381 |
| Total CCr (L/week) | 70.84 (56.01, 84.23) | 71.56 (57.30, 85.89) | 65.94 (55.98, 79.93) | .342 |
| Total Kt/V | 1.91 ± 0.56 | 1.94 ± 0.55 | 1.84 ± 0.59 | .277 |
| Peritoneal function | .138 | |||
| High/high average transport (n (%)) | 65 (34.8) | 40 (31.3) | 25 (42.4) | |
| Low/low average transport (n (%)) | 122 (65.2) | 88 (68.7) | 34 (57.6) |
BMI: body mass index; CVEs: cardiovascular events; ACEI: angiotensin converting enzyme inhibitor; ARBs: angiotensin-receptor blockers; TA-Bic: time-averaged serum bicarbonate; rGFR: residual glomerular filtration rate; CRP: C-reactive protein; LDL-C: low-density lipoprotein cholesterol; HDL-C: high-density lipoprotein cholesterol; iPTH: intact parathyroid hormone; D/P Cr: dialysate/plasma creatinine ratio; CCr: creatinine clearance rate; Kt/V: urea clearance index; A: no cardiovascular events group; B: cardiovascular events group.
Values are presented as means ± SD or interquartile range or percentages.
3.2. The relationship between serum bicarbonate levels and CVEs
The patients were divided into three groups according to TA-Bic (mmol/L) tertiles: lower tertile, <23.5; middle tertile, 23.5–25.1; and upper tertile, >25.1. Significant differences between the three groups were observed in the parameters during follow-up period, including smoking status, history of CVEs, SBP, DBP, medications (including ACEI/ARBs, beta blockers, and calcium channel blockers), albumin, and alkaline phosphatase (Table 2). However, there was a significant difference in the proportion of PD patients who developed new CVEs during follow-up period (lower vs. middle vs. upper: 50.0 vs. 36.5 vs. 8.1%, p < .001). The Kaplan–Meier survival analysis and log-rank test revealed that the cumulative patient survival rate was significantly different among three groups (Figure 2).
Table 2.
Baseline characteristics of study subjects stratified by time-averaged serum bicarbonate levels.
| Variables | TA-Bic < 23.5 (n = 62) | 23.5 ≤ TA-Bic ≤ 25.1 (n = 63) | TA-Bic > 25.1 (n = 62) | p Value |
|---|---|---|---|---|
| Age (years) | 50 (41, 56) | 50 (41, 57) | 47 (35, 53) | .222 |
| Female (n (%)) | 27 (43.5) | 39 (61.9) | 37 (59.7) | .080 |
| BMI (kg/m2) | 22.28 (19.94, 24.77) | 21.19 (19.94, 23.91) | 21.30 (19.14, 22.84) | .087 |
| Smoking (n (%)) | 27 (43.5) | 15 (23.8) | 14 (22.6) | .017 |
| Diabetes (n (%)) | 14 (22.6) | 8 (12.7) | 8 (12.9) | .229 |
| Left ventricular hypertrophy (n (%)) | 29 (46.8) | 30 (47.6) | 19 (30.6) | .096 |
| History of CVEs (n (%)) | 23 (37.1) | 19 (30.2) | 8 (12.9) | .007 |
| Incidence of peritonitis (n (%)) | 16 (25.8) | 20 (31.7) | 9 (14.5) | .073 |
| Systolic blood pressure (mmHg) | 145.26 ± 15.10 | 137.52 ± 17.27 | 134.50 ± 14.24 | <.001 |
| Diastolic blood pressure (mmHg) | 93.03 ± 11.30 | 88.84 ± 11.52 | 87.42 ± 11.07 | .017 |
| ACEI/ARBs (n (%)) | 17 (27.4) | 6 (9.5) | 14 (22.6) | .034 |
| Alpha blockers (n (%)) | 9 (14.5) | 6 (9.5) | 6 (9.7) | .605 |
| Beta blockers (n (%)) | 32 (51.6) | 17 (27.0) | 29 (46.8) | .012 |
| Calcium channel blockers (n (%)) | 56 (90.3) | 45 (71.4) | 54 (87.1) | .011 |
| Creatinine (μmol/L) | 714.45 (641.70, 892.90) | 712.00 (605.75, 854.40) | 648.45 (582.10, 828.10) | .055 |
| Uric acid (μmol/L) | 440 (380, 487) | 441 (395, 489) | 429 (373, 482) | .676 |
| rGFR (mL/min/1.73 m2) | 3.24 (2.03, 5.04) | 3.55 (2.14, 5.27) | 4.06 (3.06, 5.15) | .336 |
| White blood cell count (×109/L) | 6.04 (5.10, 7.07) | 6.18 (5.10, 7.31) | 5.47 (4.65, 7.18) | .395 |
| CRP (mg/L) | 1.62 (0.80, 4.37) | 1.24 (0.70, 4.53) | 1.20 (0.60, 2.32) | .229 |
| Serum ferritin (μg/L) | 176 (94, 343) | 157 (87, 325) | 127 (59, 260) | .346 |
| Hemoglobin (g/L) | 105.32 ± 20.02 | 106.25 ± 17.78 | 108.55 ± 18.84 | .620 |
| Albumin (g/L) | 34.76 ± 4.31 | 36.20 ± 4.95 | 37.85 ± 4.07 | .001 |
| Alkaline phosphatase (U/L) | 90 (72, 107) | 77 (58, 97) | 70 (60, 89) | .002 |
| Total cholesterol (mmol/L) | 4.33 (3.92, 4.78) | 4.66 (3.94, 5.65) | 4.39 (3.83, 5.11) | .256 |
| Triglyceride (mmol/L) | 1.35 (0.94, 1.89) | 1.29 (0.93, 1.83) | 1.39 (1.04, 2.21) | .635 |
| LDL-C (mmol/L) | 2.63 (2.25, 3.03) | 2.91 (2.29, 3.62) | 2.74 (2.00, 3.28) | .180 |
| HDL-C (mmol/L) | 1.13 (0.94, 1.39) | 1.17 (0.96, 1.45) | 1.16 (1.00, 1.49) | .377 |
| Calcium (mmol/L) | 2.20 ± 0.18 | 2.26 ± 0.17 | 2.21 ± 0.16 | .136 |
| Phosphorus (mmol/L) | 1.67 (1.34, 1.92) | 1.63 (1.33, 1.97) | 1.57 (1.34, 1.85) | .874 |
| iPTH (pg/mL) | 240 (139, 338) | 221 (94, 334) | 231 (92, 407) | .730 |
| D/P Cr at 4 h | 0.63 ± 0.12 | 0.61 ± 0.14 | 0.60 ± 0.14 | .525 |
| 24-h residual urine volume (L) | 0.80 (0.50, 1.10) | 0.80 (0.60, 1.10) | 0.82 (0.50, 1.10) | .981 |
| 24-h ultrafiltration volume (L) | 0.44 (0.05, 0.65) | 0.47 (0.19, 0.70) | 0.35 (0.01, 0.76) | .381 |
| Total CCr (L/week) | 66.06 (54.24, 82.33) | 66.53 (57.59, 83.21) | 74.95 (61.19, 84.91) | .247 |
| Total Kt/V | 1.80 ± 0.57 | 1.97 ± 0.55 | 1.95 ± 0.56 | .166 |
| Peritoneal function | .874 | |||
| High/high average transport (n (%)) | 22 (35.5) | 23 (36.5) | 20 (32.3) | |
| Low/low average transport (n (%)) | 40 (64.5) | 40 (63.5) | 42 (67.7) | |
| New CVEs during follow-up (%) | 50.0 | 36.5 | 8.1 | <.001 |
BMI: body mass index; CVEs: cardiovascular events; ACEI: angiotensin converting enzyme inhibitor; ARBs: angiotensin-receptor blockers; TA-Bic: time-averaged serum bicarbonate; rGFR: residual glomerular filtration rate; CRP: C-reactive protein; LDL-C: low-density lipoprotein cholesterol; HDL-C: high-density lipoprotein cholesterol; iPTH: intact parathyroid hormone; D/P Cr: dialysate/plasma creatinine ratio; CCr: creatinine clearance rate; Kt/V: urea clearance index.
Values are presented as means ± SD or interquartile range or percentages.
Figure 2.
Kaplan–Meier’s survival analysis according to time-averaged serum bicarbonate groups.
3.3. Independent predictors for CVEs
Univariate and multivariate Cox regression analyses were also performed to recognize the factors that affected the development of CVEs. The univariate Cox regression analysis revealed that lower baseline (HR = 0.654, 95%CI 0.565–0.757, p < .001) and time-averaged (HR = 0.622, 95%CI 0.514–0.752, p < .001) serum bicarbonate levels, lower serum albumin levels (HR = 0.828, 95%CI 0.781–0.878, p < .001), higher alkaline phosphatase levels (HR = 1.010, 95%CI 1.005–1.015, p < .001), smoking status (HR = 1.812, 95%CI 1.073–3.059, p = .026), LVH (HR = 3.831, 95%CI 2.213–6.632, p < .001), and the history of CVEs (HR = 5.084, 95%CI 2.997–8.625, p < .001) during the entire follow-up period were significant risk factors for new CVEs (Table 3). The multivariate Cox regression analysis revealed that the history of CVEs (HR = 2.421, 95%CI 1.345–4.357, p = .003) and LVH (HR = 2.043, 95%CI 1.114–3.749, p = .021) during the entire follow-up period were independent risk factors for new CVEs in PD patients, while TA-Bic level (HR = 0.767, 95%CI 0.609–0.964, p = .023) and serum albumin level (HR = 0.918, 95%CI 0.856–0.985, p = .018) were protective factors. The risk of CVEs decreased by 23.3% and 8.2% for every 1 unit increase of TA-Bic (mmol/L) and serum albumin (g/L) levels, respectively (Table 3).
Table 3.
Predictive factors associated with cardiovascular events in Cox regression analysis.
| Variable | Univariate analysis |
Multivariate analysis |
||
|---|---|---|---|---|
| HR (95%CI) | p Value | HR (95%CI) | p Value | |
| Age (years) | 1.019 (0.998–1.040) | .084 | ||
| Female (n (%)) | 1.323 (0.786–2.229) | .292 | ||
| BMI (kg/m2) | 1.017 (0.930–1.111) | .712 | ||
| Smoking (n (%)) | 1.812 (1.073–3.059) | .026 | 1.058 (0.598–1.873) | .846 |
| Diabetes (n (%)) | 1.645 (0.898–3.013) | .107 | ||
| Left ventricular hypertrophy (n (%)) | 3.831 (2.213–6.632) | <.001 | 2.043 (1.114–3.749) | .021 |
| History of CVEs (n (%)) | 5.084 (2.997–8.625) | <.001 | 2.421 (1.345–4.357) | .003 |
| Incidence of peritonitis (n (%)) | 1.142 (0.648–2.011) | .647 | ||
| Systolic blood pressure (mmHg) | 1.015 (0.999–1.032) | .071 | ||
| Diastolic blood pressure (mmHg) | 1.007 (0.985–1.030) | .528 | ||
| ACEI/ARBs (n (%)) | 1.374 (0.749–2.519) | .304 | ||
| Alpha blockers (n (%)) | 1.379 (0.652–2.916) | .401 | ||
| Beta blockers (n (%)) | 0.749 (0.437–1.286) | .295 | ||
| Calcium channel blockers (n (%)) | 2.158 (0.862–5.405) | .101 | ||
| Baseline serum bicarbonate (mmol/L) | 0.654 (0.565–0.757) | <.001 | 0.895 (0.718–1.116) | .326 |
| TA-Bic (mmol/L) | 0.622 (0.514–0.752) | <.001 | 0.767 (0.609–0.964) | .023a |
| Creatinine (μmol/L) | 1.001 (0.999–1.002) | .301 | ||
| Uric acid (μmol/L) | 1.000 (0.998–1.003) | .918 | ||
| rGFR (mL/min/1.73 m2) | 0.913 (0.797–1.046) | .188 | ||
| White blood cell count (×109/L) | 1.029 (0.896–1.180) | .687 | ||
| CRP (mg/L) | 1.015 (0.988–1.044) | .284 | ||
| Serum ferritin (μg/L) | 1.000 (0.999–1.002) | .526 | ||
| Hemoglobin (g/L) | 0.990 (0.977–1.003) | .138 | ||
| Albumin (g/L) | 0.828 (0.781–0.878) | <.001 | 0.918 (0.856–0.985) | .018b |
| Alkaline phosphatase (U/L) | 1.010 (1.005–1.015) | <.001 | 0.999 (0.991–1.006) | .728 |
| Total cholesterol (mmol/L) | 1.025 (0.822–1.279) | .826 | ||
| Triglyceride (mmol/L) | 0.950 (0.701–1.288) | .743 | ||
| LDL-C (mmol/L) | 1.092 (0.850–1.405) | .490 | ||
| HDL-C (mmol/L) | 0.765 (0.408–1.435) | .404 | ||
| Calcium (mmol/L) | 2.403 (0.538–10.741) | .251 | ||
| Phosphorus (mmol/L) | 1.119 (0.640–1.960) | .693 | ||
| iPTH (pg/mL) | 0.999 (0.997–1.000) | .069 | ||
| D/P Cr at 4 h | 3.052 (0.482–19.325) | .236 | ||
| 24-h residual urine volume (L) | 0.788 (0.426–1.458) | .448 | ||
| 24-h ultrafiltration volume (L) | 0.808 (0.502–1.298) | .378 | ||
| Total CCr (L/week) | 0.996 (0.984–1.007) | .471 | ||
| Total Kt/V | 0.669 (0.419–1.068) | .092 | ||
| Peritoneal function | ||||
| High/high average transport (n (%)) | 1.331 (0.788–2.247) | .285 | ||
HR: hazard ratio; CI: credible interval; BMI: body mass index; CVEs: cardiovascular events; ACEI: angiotensin converting enzyme inhibitor; ARBs: angiotensin-receptor blockers; TA-Bic: time-averaged serum bicarbonate; rGFR: residual glomerular filtration rate; CRP: C-reactive protein; LDL-C: low-density lipoprotein cholesterol; HDL-C: high-density lipoprotein cholesterol; iPTH: intact parathyroid hormone; D/P Cr: dialysate/plasma creatinine ratio; CCr: creatinine clearance rate; Kt/V: urea clearance index.
The hazard ratio is an indicator used to measure the risk difference between different situations or groups. It generally refers to the ratio of the risk of a specific event occurring in one group to the risk of that event occurring in another group within a specific time period. The hazard ratio for continuous variables measures the relative change in the risk of a specific outcome when there is a unit change in the continuous variable.
The risk of cardiovascular events decreased by 23.3% for every 1 unit increase of TA-Bic (mmol/L).
The risk of cardiovascular events decreased by 8.2% for every 1 unit increase of serum albumin (g/L).
Subsequently, based on the maximum lambda corresponding to the selection error mean within 1 SD of the minimum (i.e., lambda = 0.0573 and log(lambda) = −1.242), four potential predictors were derived by LASSO regression from 37 variables (Supplementary Figure 3), including a history of CVEs, LVH, baseline serum bicarbonate, and serum albumin.
3.4. Prediction nomogram of CVEs
A nomogram predicting CVEs risk after initiation of PD therapy for ESKD patients was then constructed on the basis of the above mentioned four predictors (Figure 3). The score for each predictor is shown in Table 4, and the C-index of final model was 0.806 with no collinearity (shown in Supplementary Table 1), which is higher than any other prediction model (Supplementary Figure 4).
Figure 3.
Nomogram to predict 1-, 2-, and 3-year cardiovascular events-free rate for peritoneal dialysis patients. CVEs: cardiovascular events; LVH: left ventricular hypertrophy.
Table 4.
Nomogram score for each indicator.
| Variable | Levels | Point |
|---|---|---|
| Left ventricular hypertrophy | No | 0 |
| Yes | 30 | |
| Cardiovascular disease history | No | 0 |
| Yes | 50 | |
| Baseline serum bicarbonate (mmol/L) | 18 | 88 |
| 19 | 79 | |
| 20 | 70 | |
| 21 | 61 | |
| 22 | 53 | |
| 23 | 44 | |
| 24 | 35 | |
| 25 | 26 | |
| 26 | 18 | |
| 27 | 9 | |
| Albumin (g/L) | 22 | 100 |
| 24 | 93 | |
| 26 | 86 | |
| 28 | 79 | |
| 30 | 71 | |
| 32 | 64 | |
| 34 | 57 | |
| 36 | 50 | |
| 38 | 43 | |
| 40 | 36 | |
| 42 | 29 | |
| 44 | 21 | |
| 46 | 14 | |
| 48 | 7 |
3.5. Prediction nomogram performance in the dataset
The nomogram had reliable performance for predicting the CVEs, with a time-dependent AUC of 0.782 (95%CI, 0.697–0.867), 0.828 (95%CI, 0.764–0.892), and 0.875 (95%CI, 0.801–0.949) at 1, 2, and 3 years, respectively (Figure 4). Patients were classified into low-, intermediate-, and high-risk groups according to the nomogram and Kaplan–Meier’s curves suggested that PD patients in the high risk group had significantly higher risk of CVEs (p < .001) (Figure 5). The nomogram calibration plot revealed virtually ideal predictions (Supplementary Figure 5).
Figure 4.
Time-dependent ROC curves on the basis of the nomogram. ROC: receiver operator characteristic; AUC: area under the curve.
Figure 5.
Kaplan–Meier’s survival curves on the basis of the nomogram.
To further examine the robustness of the prediction model, we performed the sensitivity analysis and internal bootstrap validation in this study. Considered the possible impact of early-onset CVEs (6 months after PD onset), there was no significant difference in the pooled effect after excluding patients with early-onset CVEs, showing ideal robustness. Similar results were also found in the internal bootstrap validation (shown in Table 5).
Table 5.
Sensitivity analysis and bootstrap internal validation to test the robustness of the prediction model.
| Variable HR (95%CI) | Nomogram model (n = 187) | Late-onset CVEs (n = 178) | Bootstrap |
|---|---|---|---|
| Left ventricular hypertrophy (n (%)) | 1.965 (1.086–3.557) | 2.113 (1.113–4.011) | 1.907 (1.034–3.514) |
| History of CVEs (n (%)) | 2.435 (1.342–4.49) | 2.456 (1.299–4.643) | 2.895 (1.443–5.808) |
| Baseline serum bicarbonatea (mmol/L) | 0.817 (0.689–0.969) | 0.850 (0.709–1.020) | 0.829 (0.694–0.990) |
| Albumin (g/L)b | 0.924 (0.864–0.989) | 0.907 (0.843–0.976) | 0.927 (0.861–0.999) |
| C-index | 0.806 | 0.826 | 0.794 |
HR: hazard ratio; CI: credible interval; CVEs: cardiovascular events.
The hazard ratio is an indicator used to measure the risk difference between different situations or groups. It generally refers to the ratio of the risk of a specific event occurring in one group to the risk of that event occurring in another group within a specific time period. The hazard ratio for continuous variables measures the relative change in the risk of a specific outcome when there is a unit change in the continuous variable.
The risk of cardiovascular events was negatively correlated with baseline serum bicarbonate levels.
The risk of cardiovascular events was negatively correlated with serum albumin levels.
4. Discussion
In the current study, we developed a practical nomogram based on LVH, a history of CVEs, serum albumin, and serum bicarbonate to predict CVEs risk among incident PD patients, showing well discrimination and sufficient accuracy.
Recent studies including us found that patients with LVH or a history of CVEs are the factors mostly commonly related to the future CVEs [14,15]. Mounting evidence has reported that hypoalbuminemia is independently associated with the increased risk of CVEs in patients undergoing renal replacement therapy [15–18]. Malnutrition, protein loss, and inflammation are three underlying mechanisms by which hypoalbuminemia may lead to CVEs, and atherosclerosis of blood vessels might be an intermediate link. Notably, this study indicated that lower bicarbonate level was associated with an increased risk of a new CVE in PD patients. Several potential mechanisms may explain the association between the risk of CVEs in dialysis patients and lower bicarbonate levels. First of all, acidosis can stimulate the secretion of endothelin-1 in vivo, which binds with endothelin A (ET-A) to induce vasoconstriction and elevation of blood pressure, as well as causing a reflexive decrease in heart rate and myocardial blood supply, promoting the development of CVEs [19]. Second, it has been suggested that chronic acidosis has a crucial effect on the activation of the renin–angiotensin–aldosterone system (RAAS), which is one of the primary cause for the development of water and sodium retention, myocardial fibrosis, and ventricular remodeling [20]. Third, acidosis inhibits cardiomyocytes Na/K-ATP activity, leading to decreased myocardial contractility, which can trigger CVEs such as heart failure. Fourth, acidosis induces the release of inflammatory factors, exacerbates the inflammatory response, causes vascular endothelial dysfunction, cardiomyocyte apoptosis and necrosis, and increases the risk of CVEs [21]. Notwithstanding, the harmfulness of acidosis in dialysis patients is often ignored. The National Kidney Foundation Disease Outcomes Quality Initiative (K/DOQI) guideline recommends an alkali supplement to CAPD patients when serum bicarbonate <22 mmol/L, which is beneficial for nutritional status and bone metabolism [22]. Interestingly, the post hoc analysis of the African American Study of Kidney Disease and Hypertension (AASK) trial showed that higher levels of serum bicarbonate (>25 mmol/L) within the normal range were related to better kidney and survival outcomes. However, there is evidence of increased mortality at bicarbonate levels of <17 or >27 mmol/L [23]. Thus, whether PD patients would benefit from higher serum bicarbonate levels remains to be further addressed.
There are limited studies about prediction models for CVEs in PD patients. Our previous study on the development of predictive scoring tools was initially designed for all PD-associated peritonitis populations, and found that the elderly, a history of CVEs, culture-positive, hypoproteinemia, and high serum alkaline phosphatase levels patients were more prone to CVEs. However, the elevated alkaline phosphatase might be affected by inflammation and infections, limiting its application in the general PD population. In the present study, we constructed a nomogram with favorable discrimination and calibration ability at 1-, 2-, and 3-years for CVEs among PD patients, which allowed early identification of patients at high risk of CVEs. Moreover, the nomogram successfully classifies patients into different who are at higher risk after discharging, as the patients might benefit the most from more frequent assessments for several years after PD initiation. Therefore, the care strategy of increasing nutrition, improving hypoalbuminemia and anemia, and correcting acidosis status might be pursued more aggressively, thereby reducing the score and turning high/intermediate-risk groups into a low-risk group, then preventing CVEs in those patients.
This study establishes a predictive model for the risk of CVEs in general PD patients for the first time, and all the variables included are easily and routinely collected in clinical care. Nevertheless, several limitations to this current study should be considered. First, the optimal serum bicarbonate level in patients with PD has not been established because of the limited sample size. Second, this is a single-center retrospective study using a small sample size and no external verification was performed, limiting our exploration and comparison to the overall PD population. Third, in our study, although a causal link was not clearly established, we observed a trend toward some association between different manifestations of CVEs (heart failure/vascular disease and possibly hypertension, etc.) and serum bicarbonate. We recognize the limitations of this finding and the need for further studies to delve into the underlying mechanisms of this association. Fourth, nutritional status (subjective global assessment, dietary protein intake, and anthropometry), history of infection, some medications use, and other cardiac function assessment indicators were not examined, which seems to be potential area of bias in this study. Fifth, we only calculated the TA-Bic levels, only one measurement of other parameter at a specific time point, thus could not represent the whole dialysis period. Importantly, data from arterial gas analyses such as PCO2 and arterial pH were not available in our study. Despite these limitations, our analysis made an important contribution to physicians in determining the cardiovascular prognosis of PD patients by creating a predictive model with high accuracy.
5. Conclusions
In conclusion, this study develop a novel nomogram composed of LVH, a history of CVEs, serum albumin, and serum bicarbonate to aid physicians in estimating the risk of 1–3 years CVEs among patients starting PD therapy. Further studies are needed to externally validate the current nomogram before clinical application.
Supplementary Material
Acknowledgements
We would like to express our gratitude for the database provided by the First Affiliated Hospital of Anhui Medical University.
Funding Statement
No funding was received for this study.
Author contributions
DS Li: conceptualization; formal analysis; investigation; methodology; roles/writing – original draft; writing – review and editing; RX Liu: investigation; methodology; project administration; resources; XM Qi: investigation; visualization; YG Wu: conceptualization; supervision; validation; roles/writing – original draft; writing – review and editing.
Ethical approval
This study was performed in accordance with the Declaration of Helsinki and was approved by the Research Ethics Committee of the First Affiliated Hospital of Anhui Medical University (Ethics approval number: PJ2023-11-52).
Consent form
All participants involved in this study provided written informed consent.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability statement
The data of this study are available on reasonable request from the corresponding author.
References
- 1.Tuttle KR, Jones CR, Daratha KB, et al. Incidence of chronic kidney disease among adults with diabetes, 2015–2020. N Engl J Med. 2022;387(15):1430–1431. doi: 10.1056/NEJMc2207018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.de Jager DJ, Grootendorst DC, Jager KJ, et al. Cardiovascular and noncardiovascular mortality among patients starting dialysis. JAMA. 2009;302(16):1782–1789. doi: 10.1001/jama.2009.1488. [DOI] [PubMed] [Google Scholar]
- 3.Kraut JA, Madias NE.. Consequences and therapy of the metabolic acidosis of chronic kidney disease. Pediatr Nephrol. 2011;26(1):19–28. doi: 10.1007/s00467-010-1564-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Chang TI, Oh HJ, Kang EW, et al. A low serum bicarbonate concentration as a risk factor for mortality in peritoneal dialysis patients. PLOS One. 2013;8(12):e82912. doi: 10.1371/journal.pone.0082912. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.D’Agostino RS, Grundy S, Sullivan LM, et al. Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA. 2001;286(2):180–187. doi: 10.1001/jama.286.2.180. [DOI] [PubMed] [Google Scholar]
- 6.Huang JC, Chen SC, Su HM, et al. Performance of the Framingham risk score in patients receiving hemodialysis. Nephrology. 2013;18(7):510–515. doi: 10.1111/nep.12094. [DOI] [PubMed] [Google Scholar]
- 7.Huang DD, Li YY, Qi XM, et al. A nomogram predicts cardiovascular events in patients with peritoneal dialysis-associated peritonitis. Ren Fail. 2022;44(1):1558–1567. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Glazkova PA, Kulikov DA, Glazkov AA, et al. Reactivity of skin microcirculation as a biomarker of cardiovascular events. Pilot study. Clin Hemorheol Microcirc. 2021;78(3):247–257. doi: 10.3233/CH-201016. [DOI] [PubMed] [Google Scholar]
- 9.Zou Y, Yang M, Wang J, et al. Association of sclerostin with cardiovascular events and mortality in dialysis patients. Ren Fail. 2020;42(1):282–288. doi: 10.1080/0886022X.2020.1741386. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Sokolow M, Lyon TP.. The ventricular complex in left ventricular hypertrophy as obtained by unipolar precordial and limb leads. Am Heart J. 1949;37(2):161–186. doi: 10.1016/0002-8703(49)90562-1. [DOI] [PubMed] [Google Scholar]
- 11.Okin PM, Roman MJ, Devereux RB, et al. Electrocardiographic identification of increased left ventricular mass by simple voltage-duration products. J Am Coll Cardiol. 1995;25(2):417–423. doi: 10.1016/0735-1097(94)00371-v. [DOI] [PubMed] [Google Scholar]
- 12.Chang TI, Kang EW, Kim HW, et al. Low serum bicarbonate predicts residual renal function loss in peritoneal dialysis patients. Medicine. 2015;94(31):e1276. doi: 10.1097/MD.0000000000001276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Nolph KD, Moore HL, Prowant B, et al. Cross sectional assessment of weekly urea and creatinine clearances and indices of nutrition in continuous ambulatory peritoneal dialysis patients. Perit Dial Int. 1993;13(3):178–183. doi: 10.1177/089686089301300303. [DOI] [PubMed] [Google Scholar]
- 14.Soliman EZ, Mills KT.. Exploring the relationship between intensive blood pressure lowering, left ventricular hypertrophy, and cardiovascular outcomes. Hypertension. 2023;80(9):1843–1844. doi: 10.1161/HYPERTENSIONAHA.123.21512. [DOI] [PubMed] [Google Scholar]
- 15.Xia X, Zhao C, Luo Q, et al. Nomogram for predicting cardiovascular mortality in incident peritoneal dialysis patients: an observational study. Sci Rep. 2017;7(1):13889. doi: 10.1038/s41598-017-14489-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Querido S, Quadros BP, Silva SH, et al. Hypervolemia, hypoalbuminemia and mitral calcification as markers of cardiovascular risk in peritoneal dialysis patients. Rev Port Cardiol. 2017;36(9):599–604. doi: 10.1016/j.repc.2016.12.014. [DOI] [PubMed] [Google Scholar]
- 17.Arques S. Hypoalbuminemia as a contributor to the progression of cardiovascular diseases. Ann Thorac Surg. 2011;92(6):2305–; author reply 2305–2306.. doi: 10.1016/j.athoracsur.2011.04.060. [DOI] [PubMed] [Google Scholar]
- 18.Suliman ME, Stenvinkel P, Bárány P, et al. Hyperhomocysteinemia and its relationship to cardiovascular disease in ESRD: influence of hypoalbuminemia, malnutrition, inflammation, and diabetes mellitus. Am J Kidney Dis. 2003;41(3 Suppl. 1):S89–S95. doi: 10.1053/ajkd.2003.50093. [DOI] [PubMed] [Google Scholar]
- 19.Yan MT, Chao CT, Lin SH.. Chronic kidney disease: strategies to retard progression. Int J Mol Sci. 2021;22(18):10084. doi: 10.3390/ijms221810084. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Yanai K, Ishibashi K, Morishita Y.. Systematic review and meta-analysis of renin–angiotensin–aldosterone system blocker effects on the development of cardiovascular disease in patients with chronic kidney disease. Front Pharmacol. 2021;12:662544. doi: 10.3389/fphar.2021.662544. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Collister D, Ferguson TW, Funk SE, et al. Metabolic acidosis and cardiovascular disease in CKD. Kidney Med. 2021;3(5):753–761.e1. doi: 10.1016/j.xkme.2021.04.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kopple JD. National kidney foundation K/DOQI clinical practice guidelines for nutrition in chronic renal failure. Am J Kidney Dis. 2001;37(1 Suppl. 2):S66–S70. doi: 10.1053/ajkd.2001.20748. [DOI] [PubMed] [Google Scholar]
- 23.Bommer J, Locatelli F, Satayathum S, et al. Association of predialysis serum bicarbonate levels with risk of mortality and hospitalization in the dialysis outcomes and practice patterns study (DOPPS). Am J Kidney Dis. 2004;44(4):661–671. doi: 10.1016/S0272-6386(04)00936-9. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data of this study are available on reasonable request from the corresponding author.